S.-B. Hong1, N. Eliaz2, G.G. Leisk2, E.M.Sachs3, R.M. Latanision2, and S.M. Allen4'Harvard School of Dental Medicine, Boston, MA, andDepartment of Materials Science and Engineering,Massachusetts Institute of Technology, Cambridge, MA;2H.H. Uhlig Corrosion Laboratory, Massachusetts Institute ofTechnology, Cambridge, MA; 3Department ofMechanicalEngineering, Massachusetts Institute of Technology,Cambridge, MA; and 4corresponding author, Department ofMaterials Science and Engineering, Massachusetts Instituteof Technology, 77 Massachusetts Avenue, Room 13-5018,Cambridge, MA 02139-4307, smallen@mit.edu

J Dent Res 80(3): 860-863, 2001

ABSTRACTThree important considerations in the fabricationof customized cranio-maxillofacial prostheses aregeometric precision, material strength, andbiocompatibility. Three-dimensional printing(3DPTM) is a rapid part-fabrication process thatcan produce complex parts with high precision.The aim of this study was to design, synthesize by3DPTM, and characterize a new Ti-5Ag (wt%)alloy. Silver nitrate was found to be an appropriateinorganic binder for the Ti powder-based skeleton,and the optimum sintering parameters for fulldensification were determined. The hardness ofthe Ti-5Ag alloy was shown to be much higherthan that of a pure titanium sample. Potentio-dynamic measurements, carried out in salinesolution at body temperature, showed that the Ti-5Ag alloy had good passivation behavior, similarto that of pure titanium. It is concluded that the Ti-Ag system may be suitable for fabrication ofcustomized prostheses by 3DPTM.

KEY WORDS: prostheses, three-dimensionalprinting (3DPTM), Ti-based alloys, corrosion.

Received July 20, 2000; Last revision January 26, 2001;Accepted February 5, 2001

A New Ti-5Ag Alloyfor Customized Prosthesesby Three-dimensional Printing3DPTM)

INTRODUCTIONThis study investigated the potential for 3DPTM fabrication of customized

complex cranio-maxillofacial prostheses with the use of a new titanium-based alloy. Currently, anatomical reconstruction for the treatment ofextensive and complex anatomic bony defects in the cranio-maxillofacialstructure involves either autogenous bone graft or implantation of artificialprosthetics made from various biomaterials (i.e., alloplastic implants). Whileautogenous bone grafts provide tissue compatibility, autograft availability islimited for large defects, and secondary surgery, bone resorption, and donorsite morbidity are concerns (Binder and Kaye, 1994). Consequently, alloplasticimplants have become an integral part of craniofacial reconstruction.

Titanium is widely used for cranioplasty and oral/maxillofacial repair(Toth et al., 1988; Chandler et al., 1994). It possesses low density, goodmechanical properties (e.g., high modulus of elasticity, toughness, and fatiguestrength), and biological and chemical inertness. In addition, because offavorable thermal and magnetic properties, titanium is compatible withmagnetic resonance imaging (MRI) and computed tomography (CT) scanning.The oxide layer on the surface of titanium and its alloys provides corrosionresistance and allows for histological osseointegration (Sullivan et al., 1994).However, because osseointegration of titanium implants takes a long time(typically from 3 to 6 months), they are often coated with hydroxyapatite [HA,Ca10(P04)6(0H)2]. This bioactive ceramic is more osteoconductive than thetitanium surface and can form direct bonds with adjacent hard tissues, leadingto an earlier fixation of the implant. Yet, the long-term survivability of HA-coated implants is still controversial, since some clinical reports have indicatedfailures caused by coating resorption or fatigue.

Three-dimensional printing (3DPTM) is a process for the rapidfabrication of three-dimensional parts directly from computer models(Sachs et al., 1992a,b, 1993). An accurate three-dimensional computermodel of a customized prosthesis can be designed with the use of a set ofCT slices (Fig. 1). This Fig. shows the skull of a patient (13 yrs 7 mosold at the time of scanning) exhibiting congenital multiple defects.Although titanium hardware may not be appropriate for a growing skull,the Fig. clearly demonstrates the potential of 3DPTM for the fabrication ofcustomized cranio-maxillofacial prostheses. The prosthesis model isdigitally sliced into discrete cross-sections for 3DPTM. The 3DPTMprocess starts by the spreading of a thin layer of powdered material,which is followed by selective joining of the powder by the printing of abinder material in areas delineated by the model cross-sections. As thisprocess is repeated layer by layer, unbound powder provides temporarysupport for part overhangs, undercuts, or internal volumes. This unboundpowder is removed upon process completion, leaving the finished"green" (packed) part.

Several significant advantages of 3DPTM over other processes incustomized prosthesis fabrication include: an ability to produce complexgeometry and small dimensions with high precision (e.g., feature-sizeresolution on the order of 100 microns, which is not attainable in casting or

860

J Dent Res 80(3) 2001

millinig of titilniunim), a potenitial lorsccaling-up ill sizc andl( Irlatc witlhm1.ultipeC prirthecads; aiid adaptabilityto difTcient ma1,1teriall systeilis.inClUtling cerrtmmics and l'unctionallyctradlcll materials ( G(iMs). [Ji-fortH'iatCly, stiitahle materialsysteilis that cxhlihit the x italcollnhillatioll of strenrgth andhiocompatihility l'or prostheticapplicationis hiave vet to heidellti flcd. 'I hsI tilc obiective of thisuNork was to desigii and syinthesizehv 3][)P''I a nICW titalliumlm-basedalloy t'ollos0ed by stndy of itssuitability l'or f'abricationt ot'CiistolIliZeCC pr-osthlses (tlhl-oulghhardness testinig electrochemicalmeasurermecnts, adi surlfacC anlalysis).

MATERIALS & METHODSTitanium Powder[)ue io its oood sinterability andflowsahility, a sphelical atoinized puietitaniulil powder was selected tor this\\ ork. I'fc comimieirciaLl powsdcr (99 9'N,pul e. SuLi-cpul-C (Chmcetals, I lorharnPark N. USA ) size was 6ff I±S1Lm in

Ti-5Ag Alloy for Customized Prostheses 861

(m ~~~¾ex

Figure 1. Computer model of skull exhibiting congenital multiple defects made from CT scans sliced at1 -mm intervals (left), and computer reconstruction of the defects (right).

diamictert A scanning electi-oni nicro-scope. SI M (Xl 3() [SEM- I(i,Philips I idhoven'I'eIhe Netherlands). was used to characterize theshapes ol tl1e PpoWdCI particles ald their sui I1ee smoothICess. X-raydilfraction. XRI) (Rotaflex, Rigaku. Osaka, Japan), was used todet-ieirmie the crxystallow aphlic strLcture of the powder particles atroomI' temLIpelatire. 'I'he tlowahility of the powder was estimatedthIoulh itlcasaitiemrnt ol' its angle ol' iepose (ASTM 20)(00).

Three-dimensional Printing (3DPTM)ilnl(det-s are L scd ill I)IDP' to join thc powder particles selectivelydurilg printitic f(iuo, 1998). Ioul diil'ferenit wax- and acrylic-p)oyinrCI-hased hinder s x cie sliowin to introducc carbollcontarminatioll into the final siiteicld part (Hong, 2(0(0(). 'I'o avoidthis pioblem. inotgnllic reactive hinders weerc selected lor iuse.leccAise silxci- exiihits hitdl solubility and dil'fusivity in titaniulil2 silxer salts xx cic examiuteed as poteintial bindeis. However. sincesilver ciahonate (AgC('() vwas showni to provide unsatisfactorybindingc of' the titaniuLm1 paiticles (I long. 200()( only silver nitrate(AgN(o) has been cised in this work. A S.8-M aqueoLus solutioll ofA\(N() (Ali'., Aesart Wardl [fill. MA, LUSA) was deposited asdroplets intIo a titaniclil powdcr bed. 'lhe powder bed wassubseqiscntl 11ChatCd fotl fir at 450('f' cliicle- argon gas atmospherein all alcimnina tthe I'Ltrnace. Because the initial siniteriligtetnpetiatiLlC (i .c. the millilmuLil temperature at wlhicli the powdeistarts to siniet- hut still easily disintegrates due to weak binding).T of' titLlitnlil powxleis is higher than 5()()(' (Hong. 2(0OO)(1iii ocund poxxdclis WCI c cVCiitciaIlfy i-ciiioved easily f'lolmi the Sxiii'aceof' the hoin fd part latCi- treat lmcult at 450"(' 'I'lie mricrostrcicUire ofthe sintelred part was characterized by SEM and XR). Wecx\aluatCd tIhe meCChlllaiCal intcgrity ol'the siutei-cc parts by placingtliCII in an1 ultraso0nic bath ((I''-120) I [CO. St. Joseph. Ml, USA)for aI'a\VcllilitLitCS.

SinteringF[Ili deiisificatioin of lthe priiited samfiles was achiexvcd in a linialsinter-ilg stage. Since titaniuLml pow(iers are typically sinteltcci to ItLilldetiSity cliider VxacUiu1i at a tCm pcl at ltre highir thali I I (f) '. tlie 1iAg systemi inivolves liqUiif-pihase sinterinei to f'tll dilensi fication.Therefore, both rnininiuni porosity and cllilol-m siSrillkaeC siotihld heensLited the forilier for imprpoved mechanlical pliopcitics andfpassivation beliavior. aid the lattev lor dimensional stability I'lieshriikage resultinig from silver riedLtioi xx asW focinci to be techiciblein comparisonxwitli the overlil shriinkage duintug t'ill densificationi(I long- 2(0f(f). A set ol' expcli iiiiets xas desi,eyed to cxaltLate thieet'fects ol' both temiipei atLilc and hevatingi rate on the nimaterialporosity We measucied the numiLber- of' poires and tilicir si/es hxanalyzing SlM micrographs of' siiteuid saifiles xxithi all imaieanalysis software package (NIII Imiiage 1 .62 NIII, litliestia(LiMI),USA), followinig a standaitd procedlI-e (ASI'M, 1)9'i) I lie poicsizC wxas iound to be loglornially distributeCd 1lie fpolosity (i c. thevofliuiie traction of pores) was fociiuil to ilierCase xitli iici easiluclievitiiig rate anid decreasinig siiitel iing teilifici attilrc (I lonug 2f)(f)f.

Characterization TechniquesTIhle I'ully siiitered sampiles wxcic chalalCtCii/ed hiy 1e1CIs ofnnicioiacrh iiiess testingi electrociemi ical meiatirLciicilts atidsurl'ace aiialysis. Ior comparison, the saniie expelivieints x\crcCanried out on PLIttC titMa4iniii control saipIes. 'Fliese saipiles xUCICcut friii a 3. 175-miiim-tliick plate (Alfa Aesar, Ward llill. MANUSA), torillecl by a seiuLelice ofi hot reollinle cold rollin-t aiicianinealinig opeiratiois. Microlia-diness iiieasiLireirints (I)M-4()ffLECO. St. osephi Ml. USA) xece coilicteCd \xitli a Vilckeisindetiter at a load ol'c2f(f g anid loadinig duation ol' IS secElectr-oclieiiicaIiiicasuLreCiiicits xVele carried OUt cii spcCillic1s

J Dent Res 80(3) 2001

'' ,vm --K r..Figure 2. Typical SEM micrograph showing the binding of titaniumparticles by silver necks following sintering of the titanium powder andsilver nitrate binder at 450 C.

xnitil 1\plosdl1c(ira of aIppI OXiirsatCl\I X Cmi11 VWC Can ied OLIthotlh rest potclitial mcasUirc ilnlltS anid pOteirtiOdynlaissiCIrrCdS1-irlrICitS Onl 4 SpCCirilCIrS to CelSUIC epCataMhilitV. I JChs meCCinilcii \x as pi Cpai (Id irs thirce steps: I ii st, a copper xx ilre xx asattacihed to the sairipICyx hothi aSxilxci pailt (Ii reIst 1'. I'ullamin.

atliarir NYs lISA) aind MinutCe Ip1Ox (I I'WI)C\coiI.I)aix\crs MA. lISA). A glass tuhc xxaxS CpOXiCdl to the samiple topI cx c\ PcpOSLIIfCof' the ire to the clecti olytc. Ncxt, thIe samiplC\\xIS irrOLIi1tC(d ilm a IPO-1iris' COld1-mOU1.t IxCSil (3UCe1leCr Lake1lUffII I. SA). iaillyl aftcr heirr CureCI. thec SpcCilslreI WasmImullil \\ithl (I0)0(-grit cirirlrdinu paper. A staidardi thilcc-clcctr-odcsystcilr (Joncs 1 996) xx aS used llot the clcctr-ochieiiiicalircaIsiircirrerts. xx\ith a satuLaWtdCI calomCl clectrodce (SC' I) as a

icl'i-cilrc lectriode. arild pILic PlatiruI-i1s foil as al CoiFterclrelectrode.A Iroi0-ddaCLratCd salieIC solLtionll (0.9' NaiN MalillrckrocdtBlakcr. Paris. KY, LISA]. p1l 5.5). maistaisilned at body

Table. Experimental Results for New Ti-5Ag (wt%) Alloys Made by ThrePrinting (3DPTM) Indicating Increased Hardness and Corrosion ResistarTitanium (values given as mean + standard deviation)

Sample Fabrication

Pure TilTi -5AgTi -5Ag

Rolling and annealingSintered at 1 300 C1Sintered at 1 150 C1

Porosity' Hardnessib% VHN

01.50 + 0 736.30 + 3.30

204 + 7700 + 38429 + 26

E orr

mV

-380 + 42-210 + 35-230 + 31

(1

'] Based on image analysis of SEM micrographs with 7 fields in each sampBased on 5 microhardness measurements each in Vickers Hardness NurRest potential was measured vs. saturated calomel reference electrode arun (4 measurements each).Corrosion current density obtained from the potentiodynamic curve gerhour of immersion in saline solution (4 measurements each).

e Control sample (plate)Heating rate 2°C/min

.(

pure ''iI i-,5Ag, 1300'(1

------.... .I'i-5A-, 15()°(

() .1!

1 F-I() II1:' CurIent ( T/crn

( turrent densil} IAcFigure 3. Typical potentiodynamic curves for Ti 5Ag (wt%) alloy madeby 3DPT. and sintered at either 1 300 or 1] 50'C The equivalent curvefor the pure titanium control sample is shown for comparison The Ti-5Ag alloy sintered at 1 300"C exhibits the lowest corrosion currentdensity and the most noble corrosion potential.

tClieClti- e (- 37 ( ). wxas uisecd as the electroltc' I hle Ircstpotential. xx.S monitorllodlito1'c1 ftOI 2) lmlill hx\ Ilmleilas 0t1 an1electrochemvricral intcrl'ace (Model SI 128N6 SchfirIrheieei.Hlaliipshire. Ilngla IId ). StlbSit) LieClrtl V potCnlltrIo(d\ ll.lnlliCte1castircm1c1ltS \\x re Carried otit t a Scan ialte of .() lVi sec fIomil-20(0 ilV I S. IO to 50ff mllV m refer clrec electlo(d. IrlecCor0roSiOnI CurrInI t (ICenSitV. i \\,ISxCxtaCxdte illIdlthroi'h(L

extra polationl of thc CathIo(diC poklar iitioil cCLI\C (potC'ItitOd\ IIrZiaI iCIISLasC rei Cut ).

RESULTSSl.MNobserxatiorns ol thie as-recei\cd titalliulm p0\\xdci Sho\\Cdthat thc paarticles had hotlh a hihl_hy spherical shape and smoo10thsirlaccs. XRIZ) cxperinlenls confirll-lmledl thlat the Particles \\Cr-c of'the hexagonal close-packed (lhtp) (-TFi phase. I he aliale olItepl)OSe IllmeaSuriILedf tIIe titaIIUmpLlI)o\\(I ci \\x as 233 ! I. 5 de(.

Alter the addition ol' tle A-NO(. hiridcl and sintcrrirrl' at45(0C. the silecr redLCed F'-orii1 AgNo(. foi-rticd \xxcll-dcf'iirednecks betwxecn the titalliu111 paltiCleS tlrIrS hrildirlr thelmitogethier (Iiig. 2). XRIl) indicated tilre co-existcnce ol' a-x Ii anid

A- ir the siuteired part. hle Sinrteredl partS \\xrc

e-dimensional f'oUrid to hax e good rehaial ulcer rt.\V itllStaIldiIll \ ibl.ltiOllS ill .111 Ultl',ISOlliC' 1-thice over Pure xlol 11101oC thall a Icx lliltLItCS.

T lic hardness \ aloeS forI the I S-Ag alloij a cr (ldelisitiCltioni at t\xo (li 'f'ctllt tepliperattilCes

A/cm2 are sho\\ II in the [1ahle in eComrparisoni \x itl apiLrr' titallil.iiri conitr-ol samiple. The printed Fi-

3 + 0 3) E 7 5Ag alloy cxllibits mItcLhCI hihilel hrrlr11ess in7.8+ 0.2) E-7 comiparisonr \\x ifl tIre cotrol sami-ple. Fig. 38.9+0.5)E-7 Sl(\\S typiCall I)(tClltiOdVIylaIlliC' C'lil\CS IL)I- tilt'3.9 + 0.5) E-7 shx xeapteirl>amiciixsFote

tlhr-ce Illaterials. Ile Shapes of) tilre erri\xCS IL11f)le. tile conitr-ol saniple arid IfOr the I i-Ag a;lo\nber units. silitered at 1 30)0(C 1ac Similar. althoiri(th} thefher 20-minute latter cxliihits a less acti\e (i c less iregalti\e)

es

cor-rosioni Potential. The alloy sirtcirdilatreroted in first 1 5(1°C exlibits less passi\ ationl at IighI

potentials ill c(illparisolil \\ithl the ot1er tx\\0riaterials. All thlree mrateriails e\liiiht sin-Lilar

862 Hong et al.

Ti-5Ag Alloy for Customized Prostheses

corrosion current density. Typical values of the rest potentialand the corrosion current density are shown in the Table.

DISCUSSIONCharacterization of the as-received titanium powder providedseveral promising indications as to its applicability to 3DPTMfabrication. First, the small powder diameter allows forreduction of the minimum thickness of printed layer attainable.The large surface area of such powder should also provide alarge driving force for sintering. Second, the smooth surfaces ofthe particles should enhance powder flowability and, hence,printability. The flowability can also be improved by thepresence of a thin oxide layer with higher hardness and lowerfriction coefficient on the surface of the titanium powder.Finally, the angle of repose measured in this work is smallerthan 25 deg, a threshold value above which a smooth powderbed is difficult to achieve (Hong, 2000).

The aqueous solution of silver nitrate was found to be agood binder for prosthetic fabrication by 3DPTM. It formedstrong interparticle silver necks at temperatures lower thanthe initial sintering temperature of titanium powders, and didnot introduce carbon contamination. The good particlebinding provided by this binder may be related to the widegap between its melting temperature (212°C) and itsdecomposition temperature to form solid silver (444°C). Thisbinder, however, has a significant drawback that has to beconsidered. Because of the high corrosiveness of thissolution, it causes some irritation to human skin and isdestructive to the 3DPTM equipment. Thus, a safe printingprocedure has to be established. Since the conventional typeof high-speed bubble-jet printhead is not recommended foruse with silver nitrate, the use of a piezo-based printhead iscurrently being studied.

Suitable parameters (temperature, heating rate, and time atmaximum temperature) for the final-stage sintering of the Ti-5Ag alloy were determined in this work. The sample that wassintered at 1300°C for 1 hr at a heating rate of 2°C/minexhibited both fairly low porosity (1.50% + 0.73%) and anarrow distribution of pore size.

The sintered Ti-Ag alloy exhibited high hardness, adesirable property of materials for prosthesis construction. Basicdesign criteria for surgical prostheses also require that thematerial be sufficiently inert in the in vivo environment. Formost metallic biomaterials, including titanium, corrosionresistance is derived from their ability to form surface-passivefilms and to retain them under the in vivo environment. Thehalf-cell electrode potential for hydrogen evolution obtainedfrom the Nemst equation (Jones, 1996) at pH 5.5 is -0.56 V (vs.SCE). Because the corrosion potential values measured for allsamples were less negative than this value, it may be assumedthat oxygen reduction is the major cathodic reaction. It shouldbe noted that because oxygen reduction is more oxidizing thanhydrogen evolution in the saline solution, the absence of de-aeration (e.g., by nitrogen gas bubbling) in this work resulted ina relatively corrosive environment. Yet, the Ti-5Ag alloyexhibited a more noble rest potential in comparison with pureTi, probably due to the higher exchange current density forhydrogen/oxygen reduction on silver than on titanium. The

slightly more noble rest potential for the sample sintered at1300°C in comparison with the one sintered at 1150°C may beattributed to a more even distribution of silver in the former.With regard to corrosion behavior, the Ti-5Ag alloy lookspromising. However, it should be noted that corrosion resistanceis only a partial indication of biocompatibility. Cell cultureand/or animal studies should be carried out before implantationof any prosthesis into human bodies.

ACKNOWLEDGMENTSThis work was supported by the 3DPTM group at MIT and madeuse of shared facilities at MIT Center for Materials Science andEngineering (CMSE), which is supported by the NationalScience Foundation under award number DMR-9400334. Thiswork is based on a thesis approved by the Harvard School ofDental Medicine in partial fulfillment of the requirements forthe degree of Doctor of Medical Sciences in the subject of OralBiology. A preliminary report was presented at the 78thGeneral Session & Exhibition of the Intemational Associationfor Dental Research, held in Washington, DC, April 5-8, 2000.

REFERENCESASTM (1999). Standard practice for determining the inclusion or

second-phase constituent content of metals by automatic imageanalysis. ASTM E 1245-95. In: 1999 annual book of ASTMstandards. Vol. 03.01. West Conshohocken, PA: American Societyfor Testing and Materials.

ASTM (2000). Standard test method for flow rate of metal powders.ASTM B 213-97. In: Annual book of ASTM standards 2000. Vol.02.05. West Conshohocken, PA: American Society for Testingand Materials.

Binder WJ, Kaye A (1994). Reconstruction of post-traumatic andcongenital facial deformities with three-dimensional computer-assisted custom-designed implants. Plast Reconstr Surg 94:775-787.

Chandler CL, Uttley D, Archer DJ, MacVicar D (1994). Imaging aftertitanium cranioplasty. Br JNeurosurg 8:409-414.

Guo H (1998). Alloy design for three-dimensional printing ofhardenable tool materials (dissertation). Cambridge, MA:Massachusetts Institute of Technology.

Hong SB (2000). Individualized cranio-maxillo-facial prosthesisconstruction using three-dimensional printing (3DPTM) withatomized titanium powder (dissertation). Boston, MA: HarvardSchool of Dental Medicine.

Jones DA (1996). Principles and prevention of corrosion. 2nd ed.Upper Saddle River, NJ: Prentice-Hall.

Sachs E, Cima M, Bredt J, Curodeau A (1992a). CAD-casting: thedirect fabrication of ceramic shells and cores by three-dimensionalprinting. ManufRev 5:118-126.

Sachs E, Cima M, Williams P, Brancazio D, Comie J (1992b). Three-dimensional printing: rapid tooling and prototypes directly from a

CAD model. JEng Indust 114:481-488.Sachs E, Haggerty J, Williams P, Cima M (1993). Three dimensional

printing techniques. US Patent 5,204,055. April 20.Sullivan PK, Smith JF, Rozzelle AA (1994). Cranio-orbital

reconstruction: safety and image quality of metallic implants on

CT and MRI scanning. Plast Reconstr Surg 94:589-596.Toth BA, Ellis DS, Stewart WB (1988). Computer-designed prostheses

for orbito-cranial reconstruction. Plast Reconstr Surg 81:315-324.

J Dent Res 80(3) 2001 863